[ntp:questions] Garmin 18 LVC as stratum 1 trouble with pps
tom kayser
forensic at milwpc.com
Mon Oct 23 06:03:26 UTC 2006
Hi,
I have studied the ntp docs, and many googles, but I cannot solve this
particular problem:
Ntp sees my Garmin 18 LVC unit as:
peer GPS_NMEA(0)
but the accuracy of the gps unit is nowhere near the expected:
offset
0.015
I was hoping to achieve. The time servers at NIST are very stable on my
machine, much more so than the gps unit. The relevent lines in
ntp.conf:
# LinuxPPS: GPS + PPS (altered)
server 127.127.20.0 minpoll 4 prefer
fudge 127.127.20.0 time1 0.18135 flag3 1 flag2 0
frequency drift:
50.982
35.948
I don't know why there are two values in that file. I am using the LinuxPPS
patch for 2.6.17 with 2.6.17.14. I have debug for PPS configured, but I can
only get debug output with the ktimer module. Otherwise I don't see any
debug output in /var/log/messages for PPS. I have tested for any activity
between dcd and ground and the pps is there.
ntpdc> kern
pll offset: -0.003363 s
pll frequency: 50.633 ppm
maximum error: 0.021496 s
estimated error: 0.005618 s
status: 0001 pll
pll time constant: 6
precision: 1e-06 s
frequency tolerance: 512 ppm
I compiled the most recent ntp with the NMEA patch, and I checked
config.h
after running configure. Everything looked ok. I did this after I compiled,
and booted with the LinuxPPS patched kernel. Incidently, in the 2.6.17
kernel support for PPS was taken out. So, the above ntpdc kern output has no
info about PPS. I also used the 2.6.15 kernel, and that ntpdc kern showed
PPS support in the kernel. The dummy timer, module ktimer works, and I can
see all the debug info for pps.c if ktimer is loaded. I cannot see any debug
info from pps.c from ttyS0.
here are some
ntpq -c rv -p
[root at localhost brian]# /usr/sbin/ntpq -c rv -p
assID=0 status=0444 leap_none, sync_uhf_clock, 4 events, event_peer/strat_chg,
version="ntpd 4.2.2p4 at 1.1585-o Sat Oct 21 14:07:48 UTC 2006 (1)",
processor="i686", system="Linux/2.6.17.14", leap=00, stratum=1,
precision=-20, rootdelay=0.000, rootdispersion=442.395, peer=31875,
refid=GPS, reftime=c8e6cd85.2ca2a244 Mon, Oct 23 2006 0:11:33.174,
poll=6, clock=c8e6cd88.42530c0e Mon, Oct 23 2006 0:11:36.259, state=4,
offset=-0.240, frequency=50.763, jitter=2.695, noise=0.588,
stability=0.071, tai=0
remote refid st t when poll reach delay offset jitter
==============================================================================
*GPS_NMEA(0) .GPS. 0 l 3 16 37 0.000 -0.240 2.695
time.nist.gov .ACTS. 1 u 3 64 3 51.589 4.351 2.787
time-a.nist.gov .ACTS. 1 u 1 64 3 62.363 1.741 7.817
time-b.nist.gov .ACTS. 1 u 1 64 3 64.483 6.216 1.453
time-a.timefreq .ACTS. 1 u 63 64 1 52.306 4.912 0.001
[root at localhost brian]# /usr/sbin/ntpq -c rv -p
assID=0 status=0444 leap_none, sync_uhf_clock, 4 events, event_peer/strat_chg,
version="ntpd 4.2.2p4 at 1.1585-o Sat Oct 21 14:07:48 UTC 2006 (1)",
processor="i686", system="Linux/2.6.17.14", leap=00, stratum=1,
precision=-20, rootdelay=0.000, rootdispersion=33.806, peer=31875,
refid=GPS, reftime=c8e6d24c.2ef759da Mon, Oct 23 2006 0:31:56.183,
poll=10, clock=c8e6d252.bc13a557 Mon, Oct 23 2006 0:32:02.734,
state=4, offset=-11.403, frequency=50.628, jitter=18.548, noise=13.277,
stability=0.001, tai=0
remote refid st t when poll reach delay offset jitter
==============================================================================
*GPS_NMEA(0) .GPS. 0 l 6 16 377 0.000 -11.403 18.548
+time.nist.gov .ACTS. 1 u 588 1024 377 51.589 4.351 4.211
+time-a.nist.gov .ACTS. 1 u 586 1024 377 62.363 1.741 9.404
-time-b.nist.gov .ACTS. 1 u 585 1024 377 63.876 5.287 3.909
-time-a.timefreq .ACTS. 1 u 585 1024 377 51.209 7.908 1.461
[root at localhost tom]#
I have let this system run for days with no better results. It's weird: I
have one server with a Garmin 18, and no pps configured, and time1 is 0.183.
That one does better than the one with pps, but curiously time1 is 0.181 on
the machine with pps. Is that a large number for time1? What else can I
try?
Brian
More information about the questions
mailing list